exo/public
14
0
Fork 0
Code Issues 15 Pull requests Packages Projects Releases Wiki Activity

gluster: add stripped pool note

Browse source
This commit is contained in:
pedrolab 2020-10-20 23:38:28 +02:00
parent abe2295dd9
commit 96376179ad
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
infrastructure/acer30/trax5678/gluster.md
View file

@ -1,6 +1,6 @@
Thanks ramon selga for sharing your knowledge and patience
Our new gluster `vm_nvme` volume is replica 3 (2 disks can fail and it still works), as we have 4 nodes (with 2 disks each one; a total of 8), but we need to have a multiple of three. Ramon suggested to do a *raid 0* in zfs to reduce the number of disks so when we multiply it by 3, we are close to ranges 10 to 20, where gluster performs good. Loosing one disk means loosing two, but replica 3 plus NVMe disk reduces the probability of an incident.
Our new gluster `vm_nvme` volume is replica 3 (2 disks can fail and it still works), as we have 4 nodes (with 2 disks each one; a total of 8), but we need to have a multiple of three. Ramon suggested to do a *stripped pool* (which behaves similar to a *raid 0*; increase capacity of disks with no redundancy). That strategy let us to reduce the number of disks so when we multiply it by 3, we are close to ranges between 10 to 20 which are good for gluster performance. Loosing one disk means loosing two, but replica 3 plus NVMe disk reduces the probability of an incident.
## zfs raid0